Sodium corrosion resistance of translucent alumina: effect of additives and sintering conditions

(1)

Sodium corrosion resistance of translucent alumina: effect of

additives and sintering conditions

Citation for published version (APA):

With, de, G., Vrugt, P. J., & Ven, van de, A. J. C. (1985). Sodium corrosion resistance of translucent alumina: effect of additives and sintering conditions. Journal of Materials Science, 20(4), 1215-1221.

https://doi.org/10.1007/BF01026317

DOI:

10.1007/BF01026317

Document status and date: Published: 01/01/1985 Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:

openaccess@tue.nl

providing details and we will investigate your claim.

(2)

J O U R N A L O F M A T E R I A L S S C I E N C E 2 0 ( 1 9 8 5 ) 1 2 1 5 - 1 2 2 1

Sodium corrosion resistance of

translucent alumina: e f f e c t of additives

and sintering conditions

G. DE W I T H , P.J. V R U G T * , A . J . C . V A N DE V E N

Phi/ips Research Laboratories, P,O.B. 80.000, 5600 JA Eindhoven, The Nether~ands

The resistance of AI203 against corrosion by sodium in the temperature range 800 to 1000 ~ C was studied by holding samples in liquid sodium in molybdenum containers for several hundred hours. Thermodynamic calculations indicating that AI203 is unstable in contact with sodium were confirmed by testing sapphire samples. A remarkable aspect was the large anisotropy in the corrosion resistance of sapphire. The effect of the sintering atmosphere and the amount of MgO and CaO dopant in polycrystalline alumina on the corrosion resistance was considered. Vacuum sintering yielded more resistant materials than hydrogen sintering. Low firing temperatures as well as low dopant levels also proved beneficial. In particular, the absence of CaO improves the corrosion resistance considerably. Consequently, the use of this additive should be avoided.

1. Introduction

Alumina is a common arc tube material for high- pressure sodium discharge lamps. Operating temperatures typically range from about 750 ~ C at the ends of the tube to about 1250 ~ C in the middle. Important properties are translucency, strength and chemical inertness. The addition of a few hundred ppm of dopant, usually MgO, is required for proper sintering to translucency. A further improvement in translucency can be obtained by adding an additional amoum of CaO

([11,

see also [2]) usually a few tenths w t p p m . It is known, however, that CaO segregates heavily on the grain boundaries of polycrystalline alumina [3]. This segregation possibly influences the corrosion ot alumina by sodium. Furthermore, an influence of the sintering atmosphere on the corrosion resistance can be expected. A detailed picture of the corrosion phenomena is not yet available in spite of several investigations [4-6]. All these investigations were done with burning lamps. The condi. tions in these lamps are rather complicated, but the experiments do indicate that the addition of CaO has a detrimental influence. In order to

obtain more insight, corrosion experiments were done on small alumina blocks encapsulated in a molybdenum box and heated in contact with sodium for several hundred hours at elevated temperature.

2. Sodium attack

When studying the stability of alumina in contact with sodium, inter-oxide compound formation should be taken into account. The following sodium-aluminates have been described in the literature [7, 8]:

Ns A, NA, NAs_ 7 (/3"-alumina), NAg_ll (/~alumina) in which

NxAy = x N a 2 0 " y A 1 2 0 ~ .

Thermodynamic data for NA and /3-alumina are given in the literature?. These substances can be formed by reactions between liquid sodium or saturated sodium vapour and A1203 according to:

3Na + 2A1~O? -+ 3 NA + A1 + AG1 (1) 6 N a + 2A1203 ~ ~ N A l l + ~ a l + AG,I. (2) *Present address: Nederlandse Philips Bedrijven B.V. Lighting Division, 5600 MD Eindhoven, The Netherlands. t All thermodynamic data were taken from references [9-11 ].

(3)

T A B L E I Material characteristics

Material Sintering MgO CaO

P(%)

Ds~

(~-~6)

(wt ppm) (wt ppm) (~m) In 1 Vac, 1850 ~ C -- 8h 100 2 99.94 19 0.52 2 Vac, 1850 ~ C -- 8 h 100 6 99.89 26 0.47 3 Vac, 1850 ~ C -- 8 h 100 16 99.91 13 0.55 4 Vac, 1850 ~ C -- 8h 100 41 99.92 18 0.78 5 H 2, 1670 ~ 100 2 99.71 11 0.37 6 H 2, 1670 ~ 300 2 99.63 15 0.56 7 H2, 1825 ~ 100 2 99.65 5 0.99 8 H 2, 1825 ~ 300 2 99.65 2 0.62

*Vac = vacuum 10 -s torr.

H2 = moist H2 (dewpoint - 15 ~ C).

In all cases a holding time of 1 h at 1500 ~ C precedes the final sintering step. The theoretical density is assumed to be 3.986 g cm -3 [13].

For other symbols see text.

The free energy changer, A G 1 , is about - - 3.3 and - - 2 . 7 k c a l at 900 and 1000 ~ C, respectively. For the second reaction, 2xGll is a p p r o x i m a t e l y constant at - - 1 . 8 k c a l . In principle, therefore, alumina is not stable in the presence o f sodium.

3. Experimental techniques

3.1. Preparation and characterization

The materials were prepared using a commercially available starting powder:~. Details o f the powder and preparation o f the ceramics are given by de With [12]. Materials 1 to 4 w were sintered in vacuum (10 -s tort) using the m i n i m u m a m o u n t o f about 1 0 0 w t p p m M g O necessary to o b t a i n a dense, translucent ceramic [13] meanwhile varying the CaO content. F o r materials 5 and 7 this mini- m u m a m o u n t Was used again b u t now sintered in moist h y d r o g e n ( d e w p o i n t 15 ~ C) at a relatively low and high temperature, respectively. Finally, materials 6 and 8 were also fired in h y d r o g e n at these two temperatures, b u t now w i t h the 300 wt p p m MgO. F u r t h e r details are given in Table I. After polishing and etching (see [12]) the area distribution o f the grains was determined. F r o m these data the v o l u m e grain size n u m b e r distribu- tions were calculated b y means of the S a l t i k o w - J o h n s o n transformation [14]. The density was d e t e r m i n e d using Prokic's m e t h o d [ 15 ].

3.2. Corrosion experiments

The kinetics o f the reactions b e t w e e n alumina and sodium were studied b y loading a m o l y b d e n u m cup with alumina (either a b l o c k or an a m o u n t o f

powder) and an overdose o f sodium. All samples were taken from the bulk o f relatively large sintered blocks to avoid sinter skin-effects. The m o l y b d e n u m was outgassed in a vacuum o f about 10 -s torr for 1 h at 1000 ~ C. The test samples (10 to 3 0 r a m 3) were cleaned for 5 m i n at 5 0 ~ in diluted (1 : 1 vol parts) HNO3, rinsed in demi-water, and dried in air. Before entering the m o l y b d e n u m cup b o t h the bulk pieces and the powders (grain size 45 to 90/~m) were fired in air for 1 h at 1000 ~ C. After this pretreatment the sample was placed in the m o l y b d e n u m cup, 20 to 50 mg sodium was added under argon atmosphere (Po2 ~ 10-3 torr) and the cup was sealed b y welding. Heating to experimental conditions was p e r f o r m e d in an elec- trically heated tube furnace, operating under a mixture o f 75% N2 and 25% H2. The t e m p e r a t u r e o f the cup was determined and controlled within 5 ~ Experiments were performed at 900 and 1000 ~ C. Holding times were 300 and 100 h, respectively. A f t e r the heat t r e a t m e n t the cup was o p e n e d in air, and the sample was immediately immersed in p e t r o l e u m awaiting further sample preparation.

In the case o f solid pieces, the test sample was cross-sectioned and polished under p e t r o l e u m in order to avoid aqueous corrosion during preparation. The samples were examined for possible corrosion products in a scanning electron microscope (SEM) using energy dispersive analysis o f X-rays (EDAX) and electron probe microanalysis (EPMA). The experiments with the powders were primarily meant to identify the reaction products "~All thermodynamic data were taken from references [9-11 ].

Ugine Kuhlman, A 15 Z.

w Throughout this paper the materials are indicated by their number in Table I.

1216

(4)

using X-ray diffraction analyses. For comparison similar tests were also done on Verneuil-grown sappire. 82

4. Results and discussion: material characteristics

The material characteristics are given in Table I. It is clear that the density,p, is essentially constant and near to the theoretical density, although somewhat lower densities are obtained for materials 5 to 8, sintered in hydrogen. From previous experiments [12] with samples taken from the bulk of the material, it is known that no CaO and only 8% MgO is lost during vacuum sintering. For hydrogen sintering the loss is presumably less.

The mean grain size,

Dso,

ranges from 5 to 26 grn and thus is not constant. Furthermore the width of the distribution u = In (Dso/D

16)

is not constant either. Two more remarks can be made. First, it should be noted that the parameters Ds0 and e only give an overall description of the grain- size distribution since some materials showed a tendency towards a bimodal distribution (see also [12]). Second, it would be a difficult if not impos- sible task to keep these parameters really constant while changing other parameters. In the interpreta- tion of data, one should be aware of this fact.

Finally, we note that there is no agreement about the benefits of CaO addition on the translucency. Although the original patent [1] and de With [12] claim an increase in transmittance, Peelen [13] did not observe this effect.

5. Results and discussion: corrosion experiments

5,1. Sapphire

The intrinsic stability of alumina in the presence of sodium was studied using sapphire as test samples. Samples were exposed to the following experimental conditions: 800 ~ C/500h, 900 ~ C/ 300 h and 1000 ~ C/100 h, respectively in contact with sodium. Basal plane sections (indices 0 0 0 1) as well as prism-like sections (indices 110 0) were investigated. The surfaces were polished, but sawn and fracture surfaces were also used for the prism- like planes.

The experiments clearly showed that alumina is indeed unstable in all three experimental conditions considered, as indicated by the thermodynamic calculations. The rate of the corrosion reaction, however, proved to be highly dependent Industrie de Pierres Scientifiques Hrand Djevahirdjian S.A.

upon crystal orientation at the surface and, of course, upon temperature. In all three conditions of temperature and time considered, the basal plane showed very little, if any, corrosion. In contrast, the prism-like planes clearly became corroded. At 800 ~ C the corrosion was limited to the outer 5 gm after 500 h. With rising temperature, corrosion at this plane increased from about 150gin in 300h at 900~ to about 180~m in 100h at 1000 ~ C. The sawn and fracture surfaces yielded very similar penetration depths. The results for the 1000 ~ C/100h treatment are shown in Fig. 1.

The X-ray diffraction diagram of sapphire powder heated for 1Oh in contact with sodium only showed peaks that could be attributed to either a-alumina or NA. None of the peaks representative of other aluminates could be detected. That the reaction product is indeed NA, and NA only, was confirmed by line scans in the corroded areas using EPMA. No indication was found that the reaction product consisted of more than one phase, whereas the A1 concentration in the corroded layer closely resembled that of NA, using alumina as a reference. The large difference in corrosion rate noted between different crystal faces indicates that stability of alumina in the presence of pure sodium is a matter of kinetics rather than of thermodyn- amics. The fact, however, that the reaction pro- ceeds in a given direction within the crystal, the reaction product being NA, proves that AG1 is negative, as predicted by Equation 1. According to Equation 2, /3-alumina could be formed thermo- dynamically, as well, but no indication was found that it actually did at these temperatures.

From these experiments it is concluded that in the temperature range considered, formation of NA at prism-like planes is the main corrosion phenomenon when single-crystalline alumina is brought into contact with sodium.

5.2, Translucent alumina: effect of

CaO dope

A sample of material 1 ( 1 0 0 w t p p m MgO, 2wt ppm CaO) was heated for 300 h at 900 ~ C in contact with sodium (Fig. 2a), and the resulting corrosion was compared with that of sapphire after the same treatment (Fig. 1 a). This polycrystalline sample did not show any detectable corrosion, whereas the prism-plane of the single crystal clearly became corroded in the same experimental CH-1870, Monthey, Switzerland.

(5)

Figure 1 Conosion of a sapphire sample after 10 h at 1000 ~ C. (a) Electron image (left • c-axis, right [I c-axis) (b) NaKt~ image (left I c-axis, right II c-axis).

conditions. Apparently the corrosion resistance of this sintered sample is not adversely affected by the presence of g a i n boundaries, nor by the 100 w t p p m MgO added. Raising the temperature to 10000 C, however, resulted in extreme corrosion after 100 h (Fig. 2b). Within 100 h all sodium added (1 mol Na per mol AlaO3) did react with the sample, causing the outer part ( ~ 0.5 mm) of the sample to disintegrate completely, the reaction product being NA. In fact the corrosion rate far exceeded that at the prism plane of sapphire after the same treatment. No indication was found that this accelerated corrosion was due to the presence of grain boundaries. Corrosion proceeded front-wise and no preferential corrosion along grain boundaries could be detected. These experi-

1218

ments were repeated with a sample of material 2 differing from material 1 in its CaO content (6 wt ppm instead of 2). The results were analogous, indicating that such small amounts of CaO do not significantly affect the stability of the grain boundaries.

Raising the CaO content further to 1 6 w t p p m (material3) resulted in serious corrosion after 300 h at 900 ~ C in contact with sodium. Corrosion was clearly initiated at the grain boundaries. Sod- ium could be detected at the grain boundaries to a depth of at least 150 ~m below the surface. Bulk corrosion was limited to the outer grain(s) (about 30/~m). Increasing the CaO content to 41 wt ppm resulted in even worse corrosion at 900 ~ C. Grain- boundary corrosion extended at least to 1000 grn

(6)

Figure 2 Corrosion of an alumina sample containing 100 wt ppm MgO and 2 wt ppm CaO (material 1) fired in vacuum (~ 10-s tort): Electron images after (a) 300 h at 900 ~ C, and (b) 100 h at 1000 ~ C.

below the surface, whereas the outer 1 0 0 ~ ' n o f the sample was c o m p l e t e l y converted into NA (Fig. 3). The excessive corrosion noted at 1000 ~ C in the case o f samples 1 and 2 was also found with samples 3 and 4.

F r o m these experiments it can be concluded that segregation o f CaO at grain boundaries ser- iously decreases the stability o f these boundaries against p e n e t r a t i o n o f sodium at 900 ~ C. A l t h o u g h there is no conclusive evidence, the presence o f CaO at the grain boundaries p r o b a b l y leads to the f o r m a t i o n o f CaO 9 6A1203 (CA6). I f it does, the resistance o f the grain boundaries against penetration o f sodium clearly depends u p o n the stability o f this c o m p o u n d against a t t a c k b y sodium.

F r o m a t h e r m o d y n a m i c point o f view CA 6 is

about as stable as A1203 itself. The free energy change for the reaction:

3 Na + ~ CA6 ~ 2 C3N2As + 37 NA + h l (3) C = CaO, N = Na20, A = AlzO3, at 8 0 0 - 1 0 0 0 ~ is a b o u t - - 4 . S k c a l , approxi- mately equal to 2xG for Reaction 1. The presence o f C A 6 at the grain boundaries evidently does n o t decrease the chemical stability o f the alumina. There is, however, a large structural similarity b e t w e e n CA6 and/3-alumina. The high m o b i l i t y o f sodium ions in this latter material is well known. A m o r e efficient initiation o f Reaction 1 in the presence o f C A 6 is thus conceivable.*

The fact t h a t corrosion in the calcia-doped sam- pies is not limited to the grain boundaries, b u t

Figure 3 Corrosion of an alumina sample containing 100 wt ppm MgO and 42 wt ppm CaO (material 4) fired in vacuum (~ 10 -s torr) after 300 h at 900 ~ C. (a) Electron image, (b) NaKa image.

*In fact, sodium corrosion experiments on CA 6 showed that the rate of attack is much larger than for A120~, although again preferentially the grain boundaries are attacked.

(7)

Figure 4 Corrosion of an alumina sample containing 100 wt ppm MgO and 2 wt ppm CaO after 300 h at 900 ~ C (NaKe image). (a) Fired in vacuum (~ 105 torr, material 1). (b) Fired in humid hydrogen (Tale w ~ 15 ~ C, material 5).

spreads out in the bulk of the grains, indicates that the apparent stability of the CaO-free materials at

9o0~162

only

results from problems in initiating the reaction in the absence of CaO (or other substances). The experiment at 1000~ using the same material indicates that at this temperature initiation is no longer a problem, and the presence of calcia is no longer a prerequisite for corrosion to occur. However, adding calcia still promotes corrosion at 1000~ because of more rapid diffusion of sodium along the grain boundaries.

From these experiments it is clear that the addition of CaO should be omitted in order to obtain optimal corrosion resistance.

5.3. Translucent alumina: effect of

sintering conditions and MgO dope

To study the effect of sintering atmosphere on stability, the corrosion resistance of material 1 was compared with that of material 5 sintered from the same starting powder, again containing 100wt ppm MgO, using the same sintering procedure except for the atmosphere, which was humid hydrogen instead of vacuum. This sample was again heated in contact with sodium for 300 h at 900 ~ C (Fig. 4).

In contrast with the material sintered in vacuum, the hydrogen-sintered sample clearly became corroded. Mthough the material did not contain appreciable amounts of CaO, corrosion was again initiated at the grain boundaries. Sodium penetration along grain boundaries extended to at least 60 gm below the surface, whereas most grains in the outer layer of the sample corroded only par-

tially. Apparently, sintefing in humid hydrogen leaves the grain boundaries in a more reactive state than sintering in vacuum. It should be noted, however, that the final MgO content of both samples after sintering might be somewhat different. Sintering in vacuum reduced the MgO content from 100 to 9 2 w t p p m [12] and this reduction is expected to be less in the case of humid hydrogen.

The effect of changing the MgO concentration was studied, keeping the CaO concentration as low as possible ( ~ 2 wt ppm). Four materials (5 to 8) were studied containing 100 (materials 5 and 6) and 300 (materials 7 and 8) wt ppm MgO, respectively, and using two sintering conditions, i.e. sintering at 1825~ (materials 5 and 7 ) a n d 1670~ (materials 6 and 8) both in humid hydrogen atmosphere.

All four examples were heated for 300h at 900 ~ C in contact with sodium. Materials 5 and 6 with the lower MgO content ( ~ 1 0 0 w t p p m ) proved to be far more resistant than materials 7 and 8 with the higher MgO content (~ 300 wt ppm). In its turn, material 6 sintered at 1670~ proved to be more resistant than material 5 sintered at 1825 ~ C, but still less resistant than the material sintered in vacuum at 1825 ~ C, i.e. material 1. The 300 wt ppm MgO materials (materials 7 and 8) suffered from severe corrosion. The outer millimetre of these samples was converted almost completely into NA, and sodium pentrated the grain boundaries throughout the samples ( ~ 4 x 4 x 4ram3). All four samples considered clearly suffered from sodium penetration along grain boundaries.

(8)

These experiments show that, irrespective of sintering temperature, the corrosion resistance of H2-sintered alumina drops drastically if the MgO content of the sinter powder is raised from 100 to 3 0 0 w t p p m . Furthermore, sintering at 1670 ~ C resulted in at least as resistant materials as sintering at 1825 ~ C.

6. Conclusion

On thermodynamic grounds it is predicted that A1203 is not stable in contact with sodium. TbSs prediction has been confirmed by corrosion experts on sapphire and (polycrystalline) alumina at temperatures ranging from 800 to 1000 ~ C. In this temperature regime, N a 2 0 " A1203 was the only detectable corrosion product. Sapphire samples showed a large anisotropy in their corrosion resistance, the basal plane being considerably less attacked than the prism-like plane. This anisotropy plays a minor role in polycrystalline alumina because in that material grain-boundary penetration dominates. For polycrystalline alumina it is concluded from corrosion experiments on ceramics sintered in hydrogen that the addition of 1 0 0 w t p p m MgO results in more resistant materials than the addition of 3 0 0 w t p p m MgO. A slightly better relative corrosion resistance is obtained after sintering at 1670~ than after sintering at 1825 ~ C. A much better corrosion resistance is obtained by firing in vacuum; in particular, grain-boundary penetration is significantly less. On the other hand, the addition of CaO, which is done to increase the optical transmittance, is detrimental to the corrosion resistance. Since it has been shown that CaO also has a harm- ful influence on the mechanical properties of

alumina [12], the addition of CaO should be avoided.

Acknowledgement

Grateful acknowledgements are due to E. Groene- woud for the electron microscope analysis and to C. Langereis for the X-ray analysis.

References

1. B.J. HUNTING and G.A. JEUNINK, US Patent 3 846 146 (1973).

2. BP 1 252 851 (1969).

3. P . E . C . FRANKEN and A. P. GEHRING, J. Mater. Sei. 16 (1981) 384.

4. E. WYNER, J. Bluminating Eng. Soe. 8 (1979) 166,

5. P. HING, ibid. 10 (1981) 194.

6. P.R. PRUD'HOMME VAN REINE, Sci. Ceram. 12

(1984) 741.

7. N.A. TOROPOV, "Handbook of Phase Diagrams of Silicate Systems", Vol, 1 (Israel Programme for Science, Jerusalem, 1972).

8. R.G. DE VRIES and W. L. ROTH, J. Amer. Ceram. Soc. 52 (1969) 367.

9. D.R. STULL and H. PROPHET, "JANAF Thermo- chemical Tables", NSRDS-NBS 37, 2nd edition (US Department o f Commerce, Washington, 1971).

10. J.T. KUMMER, Progr. Solid State Chem. 7 (1972)

141.

11. i. BARIN, o. KNACKE and O. KUBASCHEWSKI,

"Thermochemical Properties of Inorganic Sub- stances" (Springer Verlag, Berlin, 1977).

12. G. DE WITH, J. Mater. Sei. 19 (1984) 2195. 13. J. G, J. PEELEN, Thesis, University of Technology,

Eindhoven (1977).

14. H.E. EXNER, Int. Metall. Rev. 17 (1972) 25,

15. D. PROKIC, J, Phys, D. Appl, Phys, 7 (1974) 1873.

Received 30 April